itch.io

I've been building a day trading simulator solo in Godot 4.6, and I've hit the point where I need other people's hands on it rather than just my own.

The idea is narrow. Load a real recorded market day, step it forward one candle at a time, and trade it by hand without ever seeing what comes next. Not a backtester, not a paper account — closer to a replay you play.

WHERE THE BUILD IS

Working:

• Bar replay from 0.1x to 4x, with pause and single-step
• Manual long/short at the price on screen; position, equity, open P&L and worst drawdown update every candle
• 1D / 30m / 15m / 5m / 1m, and switching timeframe mid-session keeps your position and still won't reveal the next candle
• VWAP, EMA 9/20, SMA 50, Bollinger, RSI, MACD, supply/demand zones, prior-day high/low/close
• End-of-session readout: win rate, average win, average loss, profit factor
• A strategy lab that tests a rule and reports sample size, result after costs, worst drawdown, and whether it beat a random control

Not done:

• Challenge system is a stub
• One symbol loaded (AAPL)
• Fills happen at the price you clicked — no spread or slippage model yet, so results come out kinder than a real broker's
• UI gets rough at small window sizes
• Desktop only in practice; a phone will fight the interface

TWO THINGS I GOT WRONG

The strategy lab shipped with lookahead bias in its first version. A tool I built specifically to catch flattering results was flattering me. Fixed it, and a chunk of results I liked stopped being results.

The second one I found last week. Hitting Start Replay jumped straight to a zoomed-in view with no history behind it, because replay was starting from the first bar in the buffer instead of that day's session open. It now finds the session open for the day you're actually viewing, so the pre-market context is still on screen when you start.

WHAT I'D LIKE TESTED

Anything is useful, but these are where I'm least confident:

• Switch timeframes mid-replay and try to make it show you a candle it shouldn't. The whole thing rests on that one rule, so if you break it I want to know.
• Open a position, then change timeframe or jump to another day, and see whether the position and P&L survive intact.
• Run the speed control from 0.1x up to 4x and back down. Does it stall, skip bars, or drift out of sync?
• Small windows and odd aspect ratios. I've mostly tested at 1280x800.
• Load time in the browser. It's a WASM build and I'd like to know what it feels like on a slower connection.

It runs in the browser, free, nothing to install and no signup: https://propdesktrading.itch.io/propdesktrading

If it's not your thing, no hard feelings. If you do find something, drop it here or in the comments on the page and I'll reply.

NEXT DEVLOG

More symbols beyond AAPL, and a first pass at spread and slippage so the fills stop being generous. I'll post progress here as it lands.

Hey all. I've been building this solo in Godot 4.6 and there's finally a browser

build worth putting in front of people.

Play it in the browser: https://propdesktrading.itch.io/propdesktrading

What it is. You load a real recorded trading day and step through it bar by bar,

anywhere from a tenth speed to 4x. You trade it by hand as it forms. You're filled

at the price on screen when you click, and you don't get to see the next candle —

which is the entire point. Equity, P&L and max drawdown move with every bar.

Why I built it. Paper trading runs at real-world speed, so a year of practice takes

a year. A backtester compresses that same year into a millisecond and hands you a

number, but you never actually make a decision, so you never get better at making

one. I wanted the repetition of a backtest with the decision-making of live trading,

and I couldn't find it, so here we are.

What's in the demo, honestly. Bar replay, manual long and short, five timeframes

that hold your place when you switch, VWAP / EMA / Bollinger / RSI / MACD /

supply-demand zones, and a strategy lab. What isn't done: the challenge system is

bare, there's one symbol loaded, and the UI still has rough edges. It's an early

build and it plays like one.

Free, no signup, nothing to install. First load pulls about 40 MB and then caches.

What I'd actually like feedback on: whether stepping through bars is still

interesting ten minutes in, or whether it turns into a clicking chore. That's the

thing I can't judge anymore — I've been staring at it too long. If it gets boring,

I'd rather know exactly where.

Thanks for reading.